Praison AI

OPEN SOURCE2M+ DOWNLOADSBUILT IN CAMBRIDGE
For banks, insurers and hospitals

Private AI that earns trust
one task at a time.

PraisonAI helps banks, insurers and hospitals run AI agents on their own computers, even fully offline, so their private data never leaves the building.

Talk to usUse the open-source framework
YOUR BUILDINGYOUR DATAYOUR OWN AIRUNS FULLY OFFLINE
YOUR DATASTAYS INSIDE THE BUILDING
The problem

Rented AI means your data leaves the building.

Most AI agents send your data to someone else's servers. Banks, insurers and hospitals can't do that with customer or patient records, so their AI work waits on approvals that take months.

  1. 01
    Data leaves

    A rented model sees every record you send it.

  2. 02
    Approvals stall

    Compliance teams want proof before they trust AI with their data.

  3. 03
    Fine-tuning is ending

    OpenAI takes no new self-serve fine-tuning jobs after 6 January 2027.

Early access

How it works.

Each task moves to your own AI only after it passes a test. If quality slips, it moves back automatically, and every step is recorded for regulators.

The open-source framework is live today. The handover is in early access, and we are inviting our first design partners.

  1. 01
    Watch

    Records how each task is done today.

  2. 02
    Test

    Turns that real work into tests.

  3. 03
    Hand over

    A task moves to your own AI only when it passes.

  4. 04
    Roll back

    If quality drops, the task moves back automatically.

  5. 05
    Record

    Every handover leaves a record your regulator can check.

WATCHTESTHAND OVERRECORD: EVERY STEP, FOR YOUR REGULATORROLL BACK
EARLY ACCESSONE TASK AT A TIME
Early access

See it in two minutes.

The open-source framework is live today. The handover shown here is in early access.

EARLY ACCESSTWO-MINUTE DEMO
Built for regulated teams

Runs inside your walls.

Agents run on your own computers with your own models, such as Llama, Qwen and Gemma through Ollama or vLLM. They keep working with no internet connection.

  1. 01
    Banks

    Customer and transaction data stays under your own controls.

  2. 02
    Insurers

    Claims and policy records stay on your own systems.

  3. 03
    Hospitals

    Patient records never leave the building.

01

Give agents a purpose.

Connect models and tools to turn an idea into an agent that can act.

02

Let them work together.

Build multi-agent workflows for research, planning and execution.

03

Keep what matters.

Bring memory and your own knowledge into the workflow with RAG.

Get started

One command.

This installs PraisonAI. When it finishes, run praisonai setup to add a provider key. With a local model such as Ollama, no key is needed.

curl -fsSL https://raw.githubusercontent.com/MervinPraison/PraisonAI/main/install.sh | sh
THE INSTALLERTHEN PRAISONAI SETUP
Advanced installs
Python
pip install praisonaiagents
JavaScript
npm install praisonai
YAML
pip install praisonai
praisonai agents.yaml
Key by hand
export OPENAI_API_KEY="your-key"

Installation notes in the docs

The first job

Research, then a finished table.

The repository opens on this agent. It takes a question, does the research, and returns a table. This is the first job. A larger one is an Agent Team or an Agent Flow.

  1. 01
    One instruction

    Tell the agent it is a senior analyst.

  2. 02
    One question

    Ask for the research, and for a table back.

  3. 03
    The work

    A finished table, ready for the next draft or the next agent.

from praisonai import Agent

agent = Agent(instructions="You are a senior data analyst.")
agent.start("Analyze the top 3 tech trends of 2026 and format as a markdown table.")
MEET YOUR FIRST AGENTUNDER A MINUTE
For people

A 24/7 agent.

An agent can answer in the chat people already use, and a scheduler can keep the job running. The dashboard stays on your machine.

See the bot ↗
  1. 01
    Four chats

    Telegram, Discord, Slack, and WhatsApp. Add a channel from the dashboard.

  2. 02
    Memory and knowledge

    A support reply can use what the agent remembers and what it has read.

  3. 03
    On a schedule

    An agent scheduler runs the job on a clock.

Short courses

Agent University.

Each course teaches one idea. The example is on the page, and you run the agent yourself. No account is required.

Open Agent University ↗
  1. 01Agents

    Give one agent a job, check the answer, then run it on your computer.

  2. 02AgentFlow

    Pass one job through two steps, a choice, parallel work, and a loop.

  3. 03RAG

    Find the right section of a note, then answer from that section.

  4. 04Memory

    Save a fact about a person, recall it, and keep it apart from a note.

  5. 05Skills

    Load a folder of instructions only when the job matches.

  6. 06Bot

    Run the same agent in a chat app, then schedule one repeating job.

  7. 07MCP

    Attach a tool server and call one MCP tool end-to-end.

  8. 08YAML

    Define one agent in a file, validate it, then run it with the CLI.

  9. 09Deploy

    Put one agent behind a web address, call it, then run it in a container.

  10. 10Observe

    Turn on telemetry, read token cost, and send one agent run to a trace backend.

  11. 11Security

    Block unsafe output, approve tool runs, and keep secrets out of source.

  12. 12AI basics

    Pick a model, set a key safely, read tokens, and run one agent question.

We’re inviting our first design partners. Talk to us about early access. The framework stays MIT licensed.

2M+ Downloads · 9.3K Stars

The open-source framework has over 2M downloads, 9.3K stars on GitHub and 600+ releases in the last 12 months.

View on GitHub